Expert's Panel

A collection of timely and important topics posed to our panel of experts.

What global trends do you see in the SEM industry?

 (ep1080)
1. Consolidation and more consolidation. A lot of companies have not been able to keep up with the R & D required, particularly with the issues relative to 300mm, and as a result what you would refer to as the old "safe buy" syndrome has customers looking to the big, established and reliable names. In early years of the industry people were buying $50,000 – $100,000 tools. It was big bucks at the time, but not real serious dollars. Today they may not care what they pay for it, but they want it to work properly and you to be responsible for it. In the front-end there are only 3 consortia that do the bulk of the business. I was with a high-level Intel manager the other day and he asked me if I knew of any way we could help a supplier get purchased by someone else. They have great technology, but Intel didn't feel they could afford to do business with them. That feeling is becoming more and more prevalent these days. Part of every small company's planning should be to have an opportunity to be acquired. I know a company right now that has one of the finest seed layer processes relative to the copper technology. It far surpasses everyone else. The CEO absolutely insists it is going to take over the world. I don't believe they are going to go anywhere unless they hook up with a major player. Look at the collaboration of end-users. They are also linking up because the stakes are so high today.
2. Trends for the semiconductor industry are several. One, they are going to spread out risk. People have learned with the earthquake in Taiwan, the economy in Korea, the threat of wars they need to do just that. The fabs are going to be spread out in different geographic area. Second, there is a huge megatrend toward "fabless" semiconductor companies and wafer foundries. Companies have learned that the real value is at the pin level—where the pin meets the board. If a company can design, market and application engineer a chip, they don't need or want to be in the business of building the chip. It is too expensive and there are plenty of companies that want to acquire their business to build the chips. The third megatrend is continued more "shrinks" and probably larger wafers. AMD in building a 1-gigabit memory chip would be constrained if they were building at .25, .18, or .15. they will have to be at .11 or .10 microns to be able to have that switching time. Another trend is larger die sizes and "system-on-a-chip." There will be memories, RAMs, ROMs, PRAMs and all kinds of stuff on that one chip. Another trend to watch is going to all digital. I wouldn't be surprised if we see some trend to mixed-signal chips. The last trend is ecommerce, internet, cyberspace, cyberdata.
3. One more trend I would see is a cap on revenue growth for the whole industry. The decreasing cost per bit and the increasing opportunities for doing something nifty will have us running a little ahead on cost per bit, so that everything starts coming down.
4. There is one other trend that is not developed yet, but it could. If you combine the trend of fabless semiconductor houses and 12" wafers. The low level of efficiency that captive fabs tolerate will no longer be acceptable with the foundries. I'm referring here to the use of capital equipment. The numbers of equipment units sold, and probably revenues, will head downward at some point.
5. The scenario won't happen this way, but picture one day someone flips a switch and all new equipment goes to 300mm. With shrinks and new technology you have about 21/2 times the area real estate productivity. If you do the math, forgetting the cost for a moment, you find out that instead of building two fabs they will only need one. About 60% of the unit volume is gone. It won't be a square wave, but it will happen.
6. Some other thoughts on the foundry situation. The model for the big commodity manufacturers is that they want to farm out 50% of their output. They will farm it out to 2 -3 different foundries that will be breaking their necks competing to get that business. It is also politically correct for companies to have a wafer fab operation in the countries to which they are selling. The foundries address that problem.
7. Look to a parallel situation. Solectron with the assembly business was able to perfect its manufacturing processes to a degree that none of their customers could match. They could always produce their customer's products cheaper, with higher quality, and faster than could be done in-house. As a result, a lot of their customers transferred assembly outside to them.
8. IA really important trend is that our growth will now be somewhat connected to the economy—that's different from the last 30 years. When the economy is up, we will be up and visa versa. That's why you have to pay close attention to the Federal Reserve and what's going on with the economy. When the economy is down, consumers postpone purchases of computers and telecommunication devices that are so important to the IDMs.
9. More value is being pushed down the chain from the systems builders to the device manufacturers, and more work is being pushed to the equipment suppliers. In turn the equipment suppliers are pushing more to their sub-contractors. The hardware and process days are gone. We are accepting more and more responsibility from our customers. They are now looking for a complete set of solutions. They want a few suppliers who do a lot more for them.
10. I have three—What we see in the outsourcing arena, call it "value added engineering and manufacturing" is clearly the way things are going because it spreads the risk. Equally the vendor brings special areas of expertise and can become an extension of the customer's engineering. Another trend is the outsourcing vendor asking the large customer to help keep them alive during the wild downswings of the industry. Consolidation is also clearly the direction—three companies into one—good size matters these days.